x.265 half the size, double the computation; so if you double again? 1/4th?

[jasong]

In case my title wasn't clear, I'm wondering about the connection between doubling the computation time for x.265 and the reduction of the size of the video.

Basically, the wiki says that if you give twice the computation time you would give to an x.264 video to an x.265 video, the result will be about half the size of the resulting x.264 video. So, then, my question is...

What happens if you keep doubling the processing time for the x.265 video? Obviously, there are diminishing returns, but does it continue to be about half the size for every doubling of processing time? Are the numbers given simply a coincidence? (the coincidence being half the time is 1/2, double the processing time is 2, and they're reciprocals)

[VBCurtis]

If the answer to your question were yes, what would happen if you gave it enormous processing time? A 1-byte video?

It's nice that trying twice as hard results in half the file size, but that specific ratio is coincidence.

[jasong]

Originally Posted by VBCurtis

If the answer to your question were yes, what would happen if you gave it enormous processing time? A 1-byte video?

It's nice that trying twice as hard results in half the file size, but that specific ratio is coincidence.

I know taking it to the extreme creates irrational conclusions, but I'm wondering if it's approximately true in the short term.

[retina]

Originally Posted by jasong

I know taking it to the extreme creates irrational conclusions, but I'm wondering if it's approximately true in the short term.

Perhaps you could so some empirical tests and plot the results on a graph. If you are not confident in analysing the results then post the graph here if you like and see how others would analyse the curve.

[Mark Rose]

Personally I'm more excited about the Daala codec.

https://xiph.org/daala/

https://www.youtube.com/watch?v=Dmho4gcRvQ4

[VictordeHolland]

The same 1080p film encoded with x264 and x265 (half the size) I find the x264 has more visual sharpnes/details. With the same bitrate, x265 looks better (obviously). So I would say its closer to 0.7-0.8 the size of x264 for the same visual sharpness on 1080p content. With 4k, UHD, 2160p (or however you want to call) it might be closer to 0.5 as there are more pixels to encode and the algorithms of x265 work more efficient.

I still mainly use x264, because my smartphone/TV/tablet have hardware accelleration for it and storage is very cheap.

[jasong]

Originally Posted by VictordeHolland

The same 1080p film encoded with x264 and x265 (half the size) I find the x264 has more visual sharpnes/details. With the same bitrate, x265 looks better (obviously). So I would say its closer to 0.7-0.8 the size of x264 for the same visual sharpness on 1080p content. With 4k, UHD, 2160p (or however you want to call) it might be closer to 0.5 as there are more pixels to encode and the algorithms of x265 work more efficient.

I still mainly use x264, because my smartphone/TV/tablet have hardware accelleration for it and storage is very cheap.

I read that x.265 was created mainly because of concerns about mobile bandwidth in the near future.

Personally, I'd love to see Wifi routers installed at the top of light poles and along highways, with maybe some insurance offered by the Federal Government so investment losses aren't huge. Although that type of thing might already exist, I'm no expert on investment losses as they relate to American taxation.

I know a big problem with my idea is crime, so I'm thinking cheap cameras on or near the poles would be a good idea as well. Any motion not on the road itself would trigger taking a picture and then sending it back to the main server, and there'd be no requirement to view any pictures unless there was peculiar internet behavior at that point.

[xilman]

Originally Posted by jasong

I read that x.265 was created mainly because of concerns about mobile bandwidth in the near future.

Personally, I'd love to see Wifi routers installed at the top of light poles and along highways, with maybe some insurance offered by the Federal Government so investment losses aren't huge. Although that type of thing might already exist, I'm no expert on investment losses as they relate to American taxation.

I know a big problem with my idea is crime, so I'm thinking cheap cameras on or near the poles would be a good idea as well. Any motion not on the road itself would trigger taking a picture and then sending it back to the main server, and there'd be no requirement to view any pictures unless there was peculiar internet behavior at that point.

That approach might work better in a society where people aren't likely to take potshots at street fittings.